GRAIL · Function papers
What does an AI-augmented data and analytics team look like?
By Johan Grönstedt · Last reviewed
The answer arrives with its meaning attached. This is a nine-page position paper, free.
A trusted answer becomes a company service, available inside the decision and carrying its definition, sources and uncertainty with it. Analysts still interpret ambiguity. The governed answer does the retrieval, calculation and tracing around them.
The working day
A day in The 2028 analytical rhythm
Most data teams still deliver reports and answer requests. In this rhythm, the trusted answer is available inside the decision with its definition, sources and uncertainty. Analysts interpret ambiguity while the governed answer handles the retrieval, calculation and tracing around them.
The sequence starts with a bounded question and moves through a certified calculation and a trace of its meaning and coverage. An analyst reviews the low-confidence or cross-domain questions. The meeting can then move from reconciling numbers to choosing action.
ASK
A manager asks a bounded question through the company's briefing layer.
CALCULATE
Queries certified models and returns the calculation.
TRACE
Returns the calculation, definition, sources and incomplete coverage.
INTERPRET
An analyst reviews low-confidence answers, questions that cross domains.
DECIDE
The meeting moves from reconciling numbers to choosing action.
The division of work
What runs, and what stays with the person
The agent layer handles the repeatable analytical work around an answer. It drafts queries, visuals, commentary and scheduled briefs against certified measures. For ad hoc work, it clarifies the question, selects approved sources, writes and tests the query, records assumptions and produces a first interpretation. It can also find duplicate measures, draft definitions, identify conflicts and propose tests.
People retain the decisions that establish meaning or turn evidence into action. They approve new measures, consequential narratives, causal claims and executive recommendations. Domain stewards approve meaning, while the central team approves implementation. Engineers retain architecture, security, release authority, schema changes, backfills and source corrections. The agent watches tests and anomalies, traces likely causes, identifies affected assets and drafts remediation.
The same split holds when the work reaches access and data-subject support. The agent maps fields to purposes, assembles evidence, checks retention rules and routes exceptions. The DPO or legal owner decides lawful basis and disputed actions. Across the operating model, the agent retrieves, calculates, compares, traces and drafts an interpretation; the person owns ambiguity, authority and judgment.
| Process | What the agent layer does | What stays with the person |
|---|---|---|
| BI reporting and dashboards | Drafts queries, visuals, commentary, scheduled briefs and threshold alerts against certified measures | Approval of new measures and consequential narratives; the action leaders choose |
| Ad hoc requests and analysis | Clarifies the question, selects approved sources, writes and tests the query, records assumptions and produces a first interpretation | Causal claims, ambiguous questions and executive recommendations |
| Metric definitions and governance | Finds duplicate measures, drafts definitions from code and documents, identifies conflicts, proposes tests and shows downstream effects | Domain stewards approve meaning; the central team approves implementation |
| Data quality and incidents | Watches tests and anomalies, traces likely causes, identifies affected assets and drafts remediation | Engineers approve schema changes, backfills and source corrections |
| Integration and platform operation | Drafts mappings, transformation code, tests, documentation and impact analysis | Engineers retain architecture, security and release authority |
The operating model
Six things GRAIL believes
These six beliefs concern the operating model, not the tools. Together they define who owns meaning, how an answer earns trust and where human judgment enters before evidence becomes action.
The report is no longer the product. The governed answer is.
A dashboard fixes the questions and leaves interpretation to whoever opens it.
The semantic layer is management infrastructure.
Natural-language access over raw tables creates confidence before it creates meaning.
Meaning belongs in the business. Assurance belongs in the centre.
A central team cannot decide what an active customer, net revenue or productive capacity means for every function.
Failed questions are the roadmap.
Every misunderstood term, missing join and stale source shows where the analytical service is incomplete.
Augment, never automate the judgment.
The agent can retrieve, calculate, compare, trace and draft an interpretation.
Start with a decision, then earn the architecture.
The first connection should answer a repeated question that matters across systems.
Get the paper
Read the full nine-page position paper on the AI-augmented data and analytics team. The answer arrives with its definition, sources and uncertainty attached.